Supplemental Material: Distant water industrial fishing in developing countries: A case study of

1,2, 3 3 Easton R. White ú,MerrillBaker-Médard,ValeriiaVakhitova,Samantha Farquhar 4, Tendro Tondrasoa Ramaharitra5

1Department of Biological Sciences, University of New Hampshire, Durham, New Hampshire, USA 2Gund Institute for Environment, University of Vermont, Burlington, Vermont, USA 3Environmental Studies Program, Middlebury College, Middlebury, Vermont, USA 4Integrated Coastal Sciences, East Carolina University, Greenville, North Carolina, USA 5Department of Natural Science, State College of Florida Manatee-Sarasota, Bradenton, Florida, USA * Corresponding author: [email protected]

Vessel characteristics

(1e+03,Inf] (55,Inf] (500,Inf] (a) (b) (950,1e+03] (c) (450,500] (900,950] (50,55] (850,900] (400,450] (800,850] (45,50] (350,400] (750,800] (700,750] (40,45] (300,350] (650,700] (600,650] (35,40] (250,300] (550,600]

(200,250] (500,550] (30,35] (450,500] Gross tonnage (150,200] (400,450] (25,30] (kw) Engine power

Vessel length (meters) Vessel (350,400] (100,150] (300,350] (20,25] (50,100] (250,300] (200,250] (15,20] (0,50] (150,200]

0.0 0.1 0.2 0.0 0.1 0.2 0.00 0.05 0.10 0.15 0.20 Fraction of total fishing effort

Figure S1: Total fishing eort (hours) versus various vessel characteristics: (a) vessel length (meters), (b) vessel tonnage, and (c) engine power

Preprint - This work has not yet been peer-reviewed S0 60000 MDG

40000

TWN

20000 JPN Total fishing effort (hours) Total KOR CHN FRA 0 MUSSYC 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year

Figure S2: Total fishing eort (hours) for each country over time.

Preprint - This work has not yet been peer-reviewed S1 Covariates

Oil Prices

300

(a) (b)

100

200

75

100 Oil price (USD)

50 Total fishing effort (hours) Total

0

Jan−14 Jan−16 Jan−18 Jan−20 50 75 100 Date Oil price (USD)

Figure S3: (a) Global oil price (USD) over time and (b) standarized fishing eort versus the global oil price.

Preprint - This work has not yet been peer-reviewed S2 Fish prices

300 9 (a) (b)

8

200

7

6 100 Fish price index (USD) Fish price index Total fishing effort (hours) Total 5

0

Jan−14 Jan−16 Jan−18 Jan−20 5 6 7 8 9 Date Fish price index (USD)

Figure S4: (a) Fish price index (USD) over time and (b) standarized fishing eort versus the fish price index.

Preprint - This work has not yet been peer-reviewed S3 Cyclone events

300 2.0 (a) (b)

1.5 200

1.0

100

0.5 Total fishing effort (hours) Total Total number of cyclone events number Total

0.0 0

Jan−14 Jan−16 Jan−18 Jan−20 0 1 2 Date Total number of cyclone events

Figure S5: (a) Total number of cyclone events over time and (b) standarized fishing eort versus the total number of cyclone events.

Preprint - This work has not yet been peer-reviewed S4 Dipole index

1.00 300

(a) (b) 0.75

200 0.50

0.25

Dipole index 100

0.00 fishing effort (hours) Total

−0.25 0

Jan−14 Jan−16 Jan−18 Jan−20 −0.25 0.00 0.25 0.50 0.75 1.00 Date Dipole index

Figure S6: (a) Dipole index over time and (b) standarized fishing eort versus the dipole index.

Preprint - This work has not yet been peer-reviewed S5 Diagnostic plots of best fitting model

Residuals vs Fitted Normal Q−Q

161162 1023 4 1611023162 2 50 0 Residuals − 3 − 150 0 50 100 150 resid. Std. Pearson −3 −2 −1 0 1 2 3

Predicted values Theoretical Quantiles . d i

s Scale−Location Residuals vs Leverage

e 161162 r 1023

1023

n 391719 2 o s r 1.0 a e

P Cook's distance − 2

. d 0.0 t S 0 50 100 150 resid. Std. Pearson 0.00 0.02 0.04 0.06

Predicted values Leverage

Figure S7: Diagnostic residual plots for the best fitting linear model described in the main text.

Alternative models

In the main text, in our model selection process, we examined the total fishing eort for 2012-2019. Vessels were added to the Global Fishing Watch databases throughout the study making it dicult to evaluate which years of data to include or which vessels. Therefore, we present three additional analyses here with response variables as: fishing eort from 2017-2019, standardized fishing eort from 2017-2019, and only looking at data for vessels that were present every year.

Preprint - This work has not yet been peer-reviewed S6 Table S1: Parameter estimates (top value) and standard errors (bottom value in parentheses) for the average of the best fitting models (AIC < 2) using a GLM framework with a Gaussian error structure. Fishing eort (from 2017-2019) is the total daily fishing eort within the Madagascar EEZ divided by the cumulative number of vessels observed. Sine and cosine terms denote seasonal variables as a function of the julian day of the year. Fishing eort (Intercept) 28643.889 ≠ (27779.515) sine 119.333úúú ≠ (14.267) cosine 326.734úúú (8.804) year 14.229 (13.771) Dipole index 359.311úúú (38.526) Fish price index (USD) 33.968úú ≠ (12.018) Oil price (USD) 7.374úúú (1.123) Cyclone event 7.736 ≠ (23.317) Num. obs. 1094

úúúp<0.001; úúp<0.01; úp<0.05

Preprint - This work has not yet been peer-reviewed S7 Table S2: Parameter estimates (top value) and standard errors (bottom value in parentheses) for the average of the best fitting models (AIC < 2) using a GLM framework with a Gaussian error structure. Fishing eort (from 2017-2019) is the total daily fishing eort within the Madagascar EEZ divided by the cumulative number of vessels observed. Sine and cosine terms denote seasonal variables as a function of the julian day of the year. Standardized fishing eort (Intercept) 301.582úú (116.688) sine 0.696úúú ≠ (0.057) cosine 1.686úúú (0.046) year 0.149úú ≠ (0.058) Dipole index 0.759úúú (0.185) Fish price index (USD) 0.007 ≠ (0.034) Oil price (USD) 0.027úúú (0.005) Cyclone event 0.453ú ≠ (0.202) Num. obs. 1094

úúúp<0.001; úúp<0.01; úp<0.05

Preprint - This work has not yet been peer-reviewed S8 Table S3: Parameter estimates (top value) and standard errors (bottom value in parentheses) for the average of the best fitting models (AIC < 2) using a GLM framework with a Gaussian error structure. Fishing eort (from 2012-2019) is the total daily fishing eort within the Madagascar EEZ for all vessels (as opposed to only the vessels seen during all nine years of the data set used in the main manuscript). Sine and cosine terms denote seasonal variables as a function of the julian day of the year. Fishing eort (Intercept) 121906.271úúú ≠ (4774.274) sine 95.190úúú ≠ (4.954) cosine 213.735úúú (4.662) year 60.584úúú (2.373) Dipole index 110.033úúú (14.962) Fish price index (USD) 10.980úú ≠ (3.936) Oil price (USD) 1.556úúú (0.199) Cyclone event 50.379ú ≠ (19.549) Log Likelihood 19061.089 ≠ AIC 38140.179 Delta 0.000 Weight 1.000 Num. obs. 2897

úúúp<0.001; úúp<0.01; úp<0.05

Preprint - This work has not yet been peer-reviewed S9 Madagascar MPA database

Table S4: Name and year started of areas in Madagascar listed as either a MPA or LMMA

Name Status.Year Manombo 1962 Lokobe 1966 Nosy Tanihely 1968 1990 Baie de Baly 1997 Masoala Marine Park 1997 Tampolo Marine Park 1997 Tanjona Marine Park 1997 Nosy Ve 1998 Sahamalaza Iles Radama 2001 Loky Manambato 2005 Complexe Zones Humides Mahavavy 2006 Kinkony Velondriake 2006 Antongil Bay 2007 Nosy Hara 2007 Ranobe Bay 2007 Ambohibola 2008 Ambondrolava Mangroves 2008 Beheloke 2008 Itampolo 2008 Maromena/Befasy 2008 Tahosoa 2008 Amboditangena 2009 Ambodivahibe 2009 Antisakivolo 2009 Fimihara/Ankaranjelita 2009 2009 Maintimbato 2009 Manjaboaka 2009 Rantohely 2009 Vohitralanana 2009 Ambodiforaha 2011 Analanjahana 2011 Aniribe 2011 Anoromby/Andreba 2011

Preprint - This work has not yet been peer-reviewed S10 Mahasoa 2011 Tanandava 2011 Teariake 2011 Tsinjoriake 2011 Vatolava 2011 Ambodimangamaro 2012 2012 Hoalampano 2012 Seranambe 2012 -Malotrandro 2013 Sainte Luce 2014 Ambato Atsinanana 2015 Ankarea 2015 Ankivonjy 2015 Elodrato 2015 Soariake 2015 Itapera 2016 Kirindy Mite 2016 Iles Barren 2017 Site Bioculturel dAntrema 2017 Nosy Ve Androka 2018

Preprint - This work has not yet been peer-reviewed S11